The epidemiology of Clostridium difficile infection-associated reactive arthritis (CDIAReA) in children: an underdiagnosed, potentially morbid condition

Abstract

Importance

The incidence of Clostridium difficile infection has increased among children. The epidemiology of pediatric C difficile infection–associated reactive arthritis is poorly understood.

Objective

To characterize the incidence, recognition, and distinguishing clinical features of pediatric C difficile infection–associated reactive arthritis among children with C difficile infection.

Design, Setting, and Participants

In this cohort and nested case-control study using electronic health records from January 1, 2004, to December 31, 2013, across 3 geographically diverse pediatric health care networks, we screened for reactive arthritis among 148 children between ages 2 and 21 years with diagnostic or procedural codes suggesting musculoskeletal disease associated with C difficile diagnosis or positive testing. We identified 26 cases with acute arthritis or tenosynovitis within 4 weeks before to 12 weeks after confirmed C difficile infection with (1) no alternative explanation for arthritis and (2) negative synovial cultures (if obtained). Network-matched C difficile–infected controls without arthritis were randomly selected at the time of cohort member C difficile infections.

Main Outcome and Measures

Incidence of C difficile infection–associated reactive arthritis was calculated based on (1) pediatric source population and (2) children with C difficile infection. Characteristics of cases and controls were compared using conditional logistic regression.

Results

Based on the cases identified within the source population of the 3 hospital networks, we estimated that C difficile infection–associated reactive arthritis incidence was 5.0 cases per million person-years (95% CI, 3.0-7.8). Reactive arthritis affected 1.4% of children with C difficile infection yearly (95% CI 0.8%-2.3%). Joint symptoms began a median of 10.5 days after initial gastrointestinal symptoms, often accompanied by fever (n = 15 [58%]) or rash (n = 14 [54%]). Only 35% of cases of C difficile infection–associated reactive arthritis were correctly diagnosed by treating health care professionals (range across centers, 0%-64%). Five affected children (19%) were treated for presumed culture-negative septic hip arthritis despite having prior postantibiotic diarrhea and/or other involved joints. Compared with controls, cases of C difficile infection–associated reactive arthritis were less likely to have underlying chronic conditions (odds ratio [OR], 0.3; 95% CI, 0.1-0.8). Although all cases had community-onset C difficile infection and fewer comorbidities, they were more likely to be treated in emergency departments and/or hospitalized (OR, 7.1; 95% CI, 1.6-31.7).

Conclusion and Relevance

C difficile infection–associated reactive arthritis is an underdiagnosed, potentially morbid reactive arthritis associated with C difficile infection occasionally misdiagnosed as septic arthritis. Given the rising incidence of pediatric C difficile infections, better recognition of its associated reactive arthritis is needed.

INTRODUCTION

Clostridium difficile infection (CDI) is an important cause of morbidity and mortality caused by a toxin-releasing anaerobic bacterium, typically following antibiotic exposure.¹ The incidence of pediatric CDI has increased dramatically over the last 2 decades, with a 12.5-fold increase in overall incidence and 80% rise in CDI-related hospitalizations.^2,3 Hospitalized children with CDI have more infectious complications and medical comorbidities, longer lengths of stay, higher hospital costs, and greater mortality compared with hospitalized children without CDI.^4,5

Aside from its well-known gastrointestinal manifestations, CDI is also associated with inflammatory reactive arthritis. Most published information about CDI-associated reactive arthritis (CDIAReA, pronounced “see diarrhea”) comes from case reports, series, and reviews, mostly in adults.^6-10 This condition presents acutely in multiple joints recruited asymmetrically, usually beginning soon after the start of antibiotic-associated diarrhea. Clostridium difficile-associated tenosynovitis has also been described.¹¹ The HLA-B27 gene has been identified in 70% of adult cases¹⁰ and may be a risk factor for prolonged or recurrent arthritis.^6,7,12-14 The few published cases of CDIAReA in children demonstrated similar clinical features as adults.^15-19 In 2 children, CDIAReA resembled septic arthritis.^16,17

While the condition is documented in the published literature, the epidemiology of CDIAReA is poorly understood, both in adult and pediatric populations. Our study aims to objectively characterize the incidence, recognition, and distinguishing clinical features of CDI-associated reactive arthritis (CDIAReA) in children infected with C difficile.

METHODS

Study design and setting

We performed a retrospective cohort and nested case-control study using the electronic health records (EHRs) of 3 large pediatric care networks encompassing three free-standing children's hospitals—the Children's Hospital of Philadelphia (Philadelphia, PA), Nemours A.I. duPont Hospital for Children (Wilmington, DE), and Nemours Children's Hospital (Orlando, FL)—and their affiliated clinics across 4 states. This study was approved by the corresponding institutional review boards.

Study population

Eligible subjects for the cohort were between ages 2 and 21 years and diagnosed with symptomatic, laboratory-confirmed CDI between January 1, 2004, and December 31, 2013. We screened for children with CDI using the International Classification of Diseases, Ninth Revision (ICD-9) code 008.45 (previously validated for CDI in hospitalized children²⁰) and/or positive test results for C difficile. We confirmed CDI by record review as having diarrhea (≥3 loose stools/day) and either a positive C difficile stool test (toxin assay or polymerase chain reaction [PCR] test, the latter used more routinely in study centers since 2011) or gross pseudomembranous colitis, per Society for Healthcare Epidemiology of America surveillance criteria.²¹

We screened for cases of CDIAReA among children in the cohort with ICD-9-CM codes or Current Procedural Terminology (CPT) codes suggesting musculoskeletal disease (eTable 1) within 4 weeks before to 12 weeks after either CDI code or positive C difficile testing. Record review was performed to confirm cases using the following definition: presence of confirmed CDI plus acute arthritis and/or tenosynovitis documented by a physician, physician-level provider, or imaging study within 4 weeks before to 12 weeks after CDI plus (1) no other apparent cause for arthritis and (2) negative synovial fluid cultures (if obtained). While we anticipated that gastrointestinal and musculoskeletal symptoms would usually occur within several weeks after CDI, the long window was designed to maximize sensitivity using the EHR. We defined arthritis as the presence of joint effusion, swelling, or 2 other signs of joint inflammation (warmth, tenderness, restricted or painful range of motion). Children meeting the above criteria were classified as having CDIAReA, irrespective of diagnoses of treating clinicians.

For comparison, we identified controls among cohort subjects with confirmed CDI but no concomitant joint pain, arthritis, or history of CDIAReA, as confirmed by medical record review. We matched each case to up to 4 controls by date of CDI diagnosis within 3 months of diagnosis and hospital network. Matching was occasionally incomplete at Nemours from 2004 to 2010, when C difficile laboratory test results were not reliably captured in discrete fields of the EHRs, limiting the pool of eligible controls.

Study variables

We described cases and controls by demographics; comorbidities; clinical features of CDI and (for cases) arthritis; treatments used for CDI and, where applicable, arthritis; health care usage; and outcomes. We defined community-onset CDI as having diarrhea beginning outside the hospital or before hospital day 3.²² We classified CDI severity as complicated (causing hypotension, shock, ileus, megacolon, or surgical intervention), severe (acute kidney injury), or mild-to-moderate (anything else).²¹ We did not use white blood count >15,000 to define severe CDI²¹ because leukocytosis may occur with CDIAReA despite mild gastrointestinal involvement. All data were collected by record review.

Data analysis

For the pediatric source population, we calculated the yearly incidence of CDIAReA and CDI and used these results to determine the incidence of CDIAReA among children diagnosed with CDI. We excluded from incidence calculations children from Nemours in 2004 to 2010 due to unreliable recording of C difficile tests. The source population consisted of children aged 2-21 years with ≥1 encounter in the hospital network in a given year who also had a prior outpatient visit (primary or subspecialty care) in the same network within the past 24 months. This definition enabled tabulation of cases among children receiving ongoing care within the respective care network, to derive population-representative incidence estimates. A similar approach was validated in surveillance of another infectious disease.²³ We tallied cases of CDI via EHR query, counting children in the source population with both ICD-9-CM code 008.45 and positive C difficile testing in the same year. This definition has greater specificity for CDI than using diagnostic codes or test results alone.²⁰ Cases of CDIAReA included in incidence calculations were confirmed by record review to be in the source population and have both CDI diagnosis code and positive test results. We examined trends in CDIAReA incidence using linear regression and adjusted for changes in CDI incidence.

We calculated descriptive statistics for cases of CDIAReA and matched controls. We evaluated risk factors and other clinical features associated with CDIAReA using conditional logistic regression to account for matching. All analyses assumed a 2-sided type I error rate of 0.05. Results were expressed with 95% confidence intervals (CI). We used Stata/IC version 12.1 (StataCorp) for all analyses.

RESULTS

Incidence and clinical features of C difficile infection-associated reactive arthritis

The overall incidence of CDIAREA was 5.0 cases (95% CI 3.2, 7.8) per million children per year (19 children with CDIAREA detected among 3.8 million person-years of follow-up). The yearly incidence of CDIAReA rose throughout the study period by 0.8 additional cases (95% CI 0.3, 1.3) per million person-years (Table 1). Among children with CDI, the incidence of CDIAReA was 1.4 cases (95% CI 0.8, 2.3) per 100 children with C difficile infections yearly.

Table 1.

Incidence of Clostridium difficile infection (CDI) and Clostridium difficile infection-associated reactive arthritis (CDIAReA) over time

Years	Incidence of CDIAReA per million person-years (95% CI)	Incidence of CDI per 100,000 person-years (95% CI)
2004-2005	0 (0, 5.9)	45.3 (38.8, 52.6)
2006-2007	2.9 (0.3, 10.3)	33.7 (28.4, 39.6)
2008-2009	3.9 (0.8, 11.4)	26.9 (22.5, 32.0)
2010-2011	3.6 (0.7, 10.6)	32.9 (28.7, 37.5)
2012-2013	7.8 (3.1, 16.0)	47.7 (43.3, 52.4)

This table shows the overall incidence of CDIAReA (with 95% confidence intervals) in the source population over time. The incidence of CDIAReA rose during the study period.

We identified 26 children with CDI who met the predefined definition of CDIAReA (Figure). There were 17 additional possible cases (excluded from analyses) with CDI-associated arthritis with other possible etiologies or arthralgias without documented arthritis. Children with presumed CDIAReA generally developed pain, swelling, limited movement, and sometimes redness in multiple joints (median 4.5 joints, interquartile range [IQR] 3, 8) (Table 2). The pattern of arthritis was most often migratory (77%), whereby inflammation would resolve in one or several joints as new joints became inflamed. CDIAReA involved large and small joints, most commonly the hip (62%) and knee (69%), as well as tendons (eTable 2). About half of affected children had concurrent fever (58%) and/or a rash (54%), often consisting of blanching red macules and/or urticarial papules. Severe children underwent HLA-B27 testing, and results were negative (normal) for all. When performed, imaging revealed articular as well as periarticular inflammation affecting the soft tissues, fasciae, muscles, and/or bone (Table 2).

Figure.

Subject selection diagram for Clostridium difficile infection-associated reactive arthritis (cases) and Clostridium difficile infection (controls)

^a ICD-9-CM/CPT codes suggesting arthritis

^b ICD-9-CM code 008.45 and/or positive C difficile testing

^c Not mutually exclusive

^d Cases were matched to up to 4 controls by network and time of CDI ± 3 months; two cases had only 2-3 matched controls

Table 2.

Characteristics of Clostridium difficile infection-associated reactive arthritis in 26 cases

Characteristic	Number (%) or median (range)
Clinical
Pattern of joint involvement
Migratory	20 (77)
Additive	3 (12)
Other	3 (12)
Number of affected joints
1-4	13 (50)
5-8	7 (27)
≥9	6 (23)
Other symptomsa
Fever	15 (58)
Rash	14 (54)
Eye redness or pain	0
Concern for septic arthritis	12 (46)
Arthrocentesisb	11 (42)
Surgical drainagec	5 (19)
Arthritis attributed to C difficile infection	9 (35)
Laboratory
Positive testing for concurrent bacterial pathogen (N=24)	0
Positive testing for concurrent non-bacterial stool pathogen (N=18)	1 (6)d
Antinuclear antibody (N=9)	4 (44)
Rheumatoid factor (N=7)	0
HLA-B27 (N=7)	0
Imaging
Any imaging studya	19 (73)
X-ray	14 (54)
Ultrasound	9 (35)
Magnetic resonance imaging	9 (35)
Bone scan	3 (12)
Computed tomography	1 (4)
Imaging findings (N=19)a
Joint effusion	14 (74)
Synovial hypertrophy and/or enhancement	8 (42)
Soft tissue edema and/or enhancement	6 (32)
Fascial edema and/or enhancement	3 (16)
Muscle edema and/or enhancement	3 (16)
Bony edema and/or enhancement	3 (16)
Timing of symptoms, median in days (range)
From last antibiotic course to MSK symptom onset	20 (4, 28)
From GI to MSK symptom onset	10.5 (1, 32)
From MSK symptom onset to presentation	1 (0, 9)
From MSK symptom onset to arthritis diagnosis	4 (0, 44)
From MSK symptom onset to resolution	15.5 (2, 50)
From CDI diagnosis to MSK symptom resolutione	14.5 (−15, 61)
From GI symptom onset to MSK resolution	25.5 (8, 76)

CDI C difficile infection, GI gastrointestinal, IQR interquartile range, MSK musculoskeletal

^aSome children may have had more than one.

^bHips in 9 cases, knee in 1 case, ankle in 1 case.

^cHips in all 5 cases.

^dOne patient had concurrent rotavirus infection.

^eArthritis symptoms resolved in one patient treated with prednisone before C difficile diagnosis.

The onset of musculoskeletal symptoms typically followed diarrheal onset by 1-2 weeks (median 10.5 days, IQR 7, 15). The median time from onset of musculoskeletal symptoms to presentation for clinical care was 1 day (IQR 0, 3), underscoring the disruptive nature of CDIAReA. In half of children, arthritis symptoms resolved within 2 weeks (IQR 6, 22.5 days) of CDI treatment initiation, although a majority (58%) also received concurrent anti-inflammatories (mostly non-steroidal). Over median follow-up of 2.5 years (IQR 0.9, 5.0), four children with CDIAReA developed recurrent joint pains (15%), only one with a confirmed new CDI. None was diagnosed with chronic arthritis or inflammatory bowel disease.

Underdiagnosis and morbidity of CDIAReA

Clinicians providing treatment specifically attributed the arthritis to CDI in only 9 cases (35% overall; 0%, 15%, and 64% at participating centers, with overlapping confidence intervals). Alternative diagnoses included reactive arthritis from unknown causes or toxic/transient synovitis (38%), septic arthritis (19%), serum sickness (12%), acute rheumatic fever (4%), and bursitis/cellulitis (4%). Non-surgical pediatric subspecialists were involved in the care of all children diagnosed with CDIAReA, but over half of children who saw such subspecialists (53%) were not diagnosed with CDIAReA by any physician.

Nearly half of cases (46%) raised clinicians’ concerns for septic arthritis during their clinical course. These children were more likely to have hip involvement than other cases (83% vs. 43%, P=0.03). In addition to having fever, severe joint pain, and difficulty bearing weight, many of these children also had laboratory findings consistent with increased inflammation: elevated inflammatory markers and elevated white blood cell counts both in the blood and synovial fluid (median 56,375 cells/ml, IQR 50,000, 103,000; median 91.5% neutrophils, IQR 90%, 93%) (Table 3, eTable 3). The 5 children diagnosed with culture-negative septic arthritis (all affecting the hip) underwent surgical drainage procedures and completed full courses of antibiotics in addition to treatment specifically for CDI. Notably, all children whose condition resembled septic arthritis had antecedent post-antibiotic diarrhea and/or prior musculoskeletal symptoms affecting other joints.

Table 3.

Characteristics of children with Clostridium difficile infection-associated reactive arthritis clinically concerning for septic arthritis

Characteristica	Treated for septic arthritis
	Yes (N=5)	No (N=7)
Preceding antibiotic exposure	5 (100)	6 (86)
Preceding gastrointestinal symptoms (e.g., diarrhea)	4 (80)	6 (86)
Days from first GI symptom to first MSK symptom, median (IQR)	11 (10, 28)	12 (3, 16)
Other symptomatic joints	5 (100)	7 (100)
Other symptomatic joints before surgical drainage	4 (80)	-
Other symptomatic joints after surgical drainage	1 (20)	-
Number of symptomatic joints, median (IQR)	2 (2, 3)	5 (4, 8)
Presence of fever	5 (100)	4 (57)
White blood count, peak, median (IQR)	15.3 (14.7, 20.7)	17.7 (16.4, 21.9)
C-reactive protein (mg/dl), peak, median (IQR)	6.9 (5.9, 14.9)	8.1 (3.8, 13.0)
Erythrocyte sedimentation rate, peak, median (IQR)	58 (57, 78)	30 (24, 50)
Synovial white blood cells (103/ml), median (IQR)	103 (86, 212)	32 (10, 50)
Synovial neutrophils (%), median (IQR)	92 (91, 93)	79 (75, 82)
Documented reason(s) for surgical interventionb
High synovial white cell count	4 (80)	-
Refusal to bear weight	3 (60)	-
Micromotion pain/tenderness	3 (60)	-
Other laboratory abnormalities	2 (40)	-
Positive synovial fluid cultures	0/5	0/6c
Misdiagnosed with concurrent osteomyelitis	2 (40)d	0
Diagnosed with Clostridium difficile infection-associated reactive arthritis	1 (20)	3 (43)
Completed full antibiotic treatment for septic arthritis	5 (100)	0

GI gastrointestinal. IQR interquartile range. MSK musculoskeletal.

^aData expressed as N (%) unless otherwise noted.

^bSome children had more than 1 documented reason for surgical intervention.

^cSynovial fluid not obtained from 1 individual.

^dBoth individuals had a migratory arthritis that followed the development of post-antibiotic diarrhea, and both had extensive extra-articular inflammation on imaging affecting the surrounding soft tissues, fascia, bone (misdiagnosed as osteomyelitis), and, in one case, muscle.

Comparison of C difficile-infected children with and without reactive arthritis

Among children with CDI, cases and controls had similar demographic characteristics (Table 4). Compared with controls, children with CDIAReA were less likely to have a chronic underlying condition (OR 0.3, 95% CI 0.1, 0.8, P=0.01) and had significantly fewer medical comorbidities (OR 0.5, 95% CI 0.3, 0.8, P=0.009). Cases did not differ from controls in their history of autoimmune comorbidities or arthritis. With regard to the CDI itself, cases were more likely to have community-onset CDI (100% cases, 75% controls, P=0.005). Otherwise, there were no differences between cases and controls in their history of preceding infections or antibiotic exposures, CDI severity, or CDI treatment.

Table 4.

Characteristics of children with Clostridium difficile infection with (cases) and without (controls) reactive arthritis

Characteristic	Cases (N=26)	Controls (N=101)	OR (95% CI)
Demographics
Age, median (IQR)	8 (6, 14)	8 (3, 13)	1.04 (0.96, 1.1)
Female sex, N (%)	13 (50)	53 (52)	1.2 (0.5, 2.8)
Race, N (%)
White	19 (73)	73 (72)	1.3 (0.5, 3.7)
Black	3 (12)	18 (18)	0.6 (0.2, 2.2)
Other	4 (15)	10 (10)	2.0 (0.5, 8.8)
Ethnicity, N (%)
Not Hispanic/Latino	23 (88)	94 (93)	0.5 (0.1, 2.4)
Hispanic/Latino	1 (4)	6 (6)	0.6 (0.1, 5.7)
Unknown	2 (8)	1 (1)	-a
Medical history
Any comorbidity,b N (%)	13 (50)	78 (77)	0.3 (0.1, 0.8)
Cardiovascular	3 (3)	6 (6)
Endocrine	2 (8)	5 (5)
Gastrointestinal	1 (4)	41 (41)
Genetic/Congenital	0	5 (5)
Hematologic	0	6 (6)
Immunologic	2 (8)	24 (24)
Neuromuscular	2 (8)	15 (15)
Oncologic	1 (4)	16 (16)
Psychiatric	3 (12)	0
Renal	1 (4)	10 (10)
Respiratory (any)	4 (15)	18 (18)
Respiratory (not asthma)	1 (4)	11 (11)
Other	2 (8)	4 (4)
Autoimmune comorbidity, N (%)	3 (12)	16 (16)	0.7 (0.2, 2.6)
Inflammatory bowel disease	1 (4)	14 (14)
Systemic rheumatic disease	1 (4)	1 (1)
Other	1 (4)	1 (1)
History of arthritis (any cause), N (%)	1 (4)	1 (1)	-a
Number of medical comorbidities, median (IQR)	0 (0, 1)	1 (1, 2)	0.5 (0.3, 0.8)
Clostridium difficile infection
Prior C difficile infection, N (%)	1 (4)	11 (11)	0.4 (0.1, 2.8)
Community-onset C difficile infection,c N (%)	26 (100)	76 (75)	-d
Severe C difficile infection, N (%)	1 (4)	9 (9)	0.4 (0.1, 3.5)
Infection in prior 2 months,b N (%)	19 (73)	68 (67)	1.3 (0.5, 3.4)
Bloodstream (including central line with fever)	0	13 (13)
Upper respiratory tract	13 (50)	27 (27)
Lower respiratory tract	1 (4)	12 (12)
Gastrointestinal	0	9 (9)
Skin and soft tissue	4 (15)	14 (14)
Urinary tract	1 (4)	6 (6)
Other	4 (15)	3 (3)
Surgical procedure in prior 2 months, N (%)	3 (12)	19 (19)	0.6 (0.2, 2.1)
Antibiotic exposure in prior 2 months,b N (%)	22 (85)	80 (79)	1.5 (0.4, 4.8)
Penicillins	2 (8)	13 (13)
Penicillins with beta-lactamase inhibitor	3 (12)	11 (11)
Penicillins, antipseudomonal	0	9 (9)
Cephalosporins	10 (38)	39 (39)
Clindamycin	6 (23)	18 (18)
Sulfonamides	2 (8)	20 (20)
Broad-spectrum gram-positive (e.g., linezolid)	0	15 (15)
Macrolides	0	7 (7)
Fluoroquinolones	0	9 (9)
Aminoglycosides	0	15 (15)
Tetracyclines	0	2 (2)
Other	3 (12)	13 (13)
Prophylactic antibiotic exposure, N (%)	2 (8)	5 (5)	1.5 (0.3, 8.3)
Days of antibiotics in last 60 days, median (IQR)	10 (7, 19)	10 (10, 21)	0.94 (0.88, 1.01)
Days from onset of GI symptoms to diagnosis of C difficile infection, median (IQR)	12.5 (6, 16)	7 (2, 14)	1.02 (0.98, 1.05)
Treatment of C difficile infection,b N (%)
Oral metronidazole	24 (92)	93 (92)	1 (0.2, 4.9)
Intravenous metronidazole	1 (4)	3 (3)	-a
Oral vancomycin	4 (15)	10 (10)	1.8 (0.5, 7.1)
Health utilization
Seen in ED and/or hospitalized (any reason), N (%)	24 (92)	65 (64)	7.1 (1.6, 31.7)
Seen in ED and/or acutely hospitalized for C difficile infection-related illnesse, N (%)	24 (92)	40 (40)	32.1 (4.2, 245)
Hospitalized during C difficile infectionb, N (%)	21 (81)	54 (53)	4.5 (1.4, 14.2)
Hospitalized for C difficile symptoms or dehydration	6 (23)	29 (29)	0.8 (0.3, 2.2)
Hospitalized for arthritis	17 (65)	0	-
Hospitalized for other reason	0	25 (25)	-

ED emergency department

^aOR not shown when the total number of affected cases and controls is <5.

^bSome patients have more than 1 category listed

^cOnset of symptoms outside of the hospital or before hospital day 3

^dOR undefined because all cases had community-onset C difficile infections; P = 0.005 by chi-square test

^eHospitalization for abdominal symptoms, dehydration, or arthritis, not including prior hospitalization leading to hospital-onset C difficile infection

Despite having fewer comorbidities and exclusively community-onset CDI, children with CDIAReA had notably higher rates of healthcare utilization compared with controls (Table 4). Nearly all cases (N=24, 92%) were seen in emergency departments and/or hospitalized for their acute CDI-related illness. Two-thirds (N=17, 65%) were hospitalized specifically for arthritis. Children with CDIAReA saw various pediatric generalists and specialists other than their primary care physicians for their acute illness (eTable 4).

DISCUSSION

To our knowledge, this is the first systematic evaluation of the clinical epidemiology of CDIAReA. CDIAReA is an underdiagnosed and often morbid condition affecting 1-2% of children with CDI. CDIAReA was sometimes identified as a reactive process (reactive arthritis, transient synovitis, or serum sickness) by clinicians, but it was attributed to CDI in a minority of cases. Complicating this issue further, CDIAReA was occasionally misdiagnosed as a septic arthritis owing to severe, isolated joint (usually hip) pain and systemic inflammation (e.g., fever, marked laboratory abnormalities). Some children with CDIAReA were treated for presumed culture-negative septic arthritis despite clinical clues suggestive of an alternative diagnosis, namely, preceding post-antibiotic diarrhea and other transiently symptomatic joints. Furthermore, nearly all affected children were seen in emergency departments and/or were hospitalized acutely even though they had fewer medical co-morbidities and presented more frequently with community-onset CDI than matched C difficile-infected comparators. These findings, along with the rising incidence of CDIAReA, suggest that better recognition of this condition is needed to treat C difficile promptly, while avoiding unnecessary, potentially harmful interventions.

Clostridium difficile is considered an uncommon cause of reactive arthritis compared with other enteric pathogens.²⁴ Over 50 cases of CDIAReA have been published in the medical literature since 1976, predominantly in adults.^6-10 Case reports reviewing other published cases have highlighted common features of CDIAReA: onset of arthritis within 1-3 weeks of gastrointestinal symptoms; frequent presence of fever; migratory involvement of multiple large and small joints (most commonly wrists, knees, and ankles in adults); and symptomatic resolution for most patients within 2 months of treatment for C difficile.^8,10,25 While disparate reported cases cannot be considered a representative sample, these clinical features are compatible with the few reported pediatric cases^15-19 and our own study's findings. In contrast with adults, children appear to have more frequent hip involvement and, associated with this, more dramatic clinical presentations that sometimes raise concerns for septic arthritis.^16,17

Several children with CDIAReA in our cohort were diagnosed with serum sickness. Generally, this immune complex-mediated disease presents within 1-2 weeks after antibiotic exposure as a symmetric arthritis of multiple joints accompanied by urticarial-like rash and angioedema. In contrast, in our cohort, CDIAReA usually presented over 2 weeks after the most recent antibiotic course with asymmetric, migratory joint inflammation and sometimes rash without angioedema.

The passing resemblance of some cases of CDIAReA to septic arthritis can pose greater challenges in management because of the potential morbidity of joint infections as well as their treatment. In some children, it can be difficult to discern whether the joint is infected, so treatment likely should be aggressive. Drainage procedures, even when performed outside the operating room, may provide symptomatic relief as well as fluid for analysis and culture.^16,17 However, for those with preceding (post-antibiotic) diarrhea and migratory or multiple-site arthritis, CDIAReA can be presumed and closely monitored. In our study, CDIAReA was self-limited, usually resolving within 1-3 weeks after treatment for CDI. Use of oral anti-inflammatories (non-steroidal, occasionally glucocorticoids) was a helpful adjunct for children with more severe or persistent symptoms. Whether prompt recognition and treatment of CDI could prevent this rheumatic complication is unclear.

We did not detect clinical risk factors for CDIAReA among those with pre-existing CDI; cases and controls had similar demographics, overall rates of antecedent infection and antibiotic exposure, and treatment of CDI upon diagnosis. While there were some group differences in specific infections and antibiotics before CDI, these differences more likely reflected the better baseline health of cases. The co-occurrence of arthritis with C difficile infection may relate to other genetic, immunologic, and microbiologic factors, as reported with other infectious triggers.^26-28 Interestingly, some children with C difficile-associated arthralgias were neutropenic with their illness, which may have prevented frank arthritis from manifesting. In contrast, those with CDIAReA resembling septic arthritis had robust neutrophilic immune responses in the blood and synovial fluid. These observations highlight the potential importance of the innate immune system in the severity of CDIAReA. Notably, C difficile toxin is a potent activator of innate immunity.^29,30 Moreover, C difficile can also stimulate cytokines important for the activity of Th17 cells,^31,32 adaptive immune cells important in the pathophysiology of chronic arthritis.³³

Our study has several strengths. We used a systematic approach to identify cases of CDIAReA along with matched controls in the EHR of several large pediatric hospital networks across 4 states. This approach allowed us to study the incidence and risk factors for this disease within a large pediatric population, yielding novel epidemiologic data on this poorly understood disease. Of note, rates of pediatric CDI in our study were compatible with rates reported elsewhere during the same period.² In addition, our case-finding method allowed us to evaluate children with CDIAReA whose diagnosis was not recognized by treating clinicians, as well as study a wider spectrum of clinical disease than might otherwise have been possible based on referral alone.

This study also has limitations. Our case-finding algorithm relied on documentation of both CDI and musculoskeletal problems within a limited window of time. It is possible that we missed cases either because of mild symptoms or incomplete documentation. We identified other children with CDI and arthralgias/arthritis suggestive of CDIAReA who did not meet our study definition because of alternative possible etiologies or the lack of documented arthritis. For this reason, we may have underestimated CDIAReA incidence, and we may have focused on children with more severe musculoskeletal disease and thus overestimated CDIAReA severity. On the other hand, because reactive arthritis is a diagnosis of exclusion, we cannot be certain that CDI triggered arthritis in all cases. Testing for alternative bacterial pathogens was not comprehensive, and 1 child also tested positive for rotavirus (not generally considered a cause of reactive arthritis). Furthermore, the apparent rise in CDIAReA incidence during the study period may have reflected other secular trends, such as improving EHR documentation or the introduction of sensitive PCR-based techniques to detect C difficile. Additionally, although we studied CDIAReA within large pediatric populations across several states, our findings may still reflect local patterns of CDI, genetic susceptibility, and practice patterns that may not generalize to other regions. In our study, few patients were tested for HLA-B27, and all those that were had negative results, limiting more systematic examination of a genetic risk factor linked to joint inflammation in response to gut microbiota imbalance.^34-36 Finally, though this is the largest epidemiologic study of CDIAReA to date, we may have been limited in statistical power to identify all clinical factors associated with this disease.

Conclusions

CDIAReA is an underdiagnosed, potentially morbid reactive arthritis associated with CDI that can occasionally present similarly to septic arthritis. A history of migratory or multiple joint pains and concurrent post-antibiotic diarrhea may be important diagnostic clues in children presenting with fever and severe joint pain. CDIAReA leads to high rates of healthcare usage even though many affected children are otherwise healthy and have community-associated CDIs. As the incidence of CDIs and CDIAReA increases in children, better recognition of CDIAReA is needed to initiate prompt treatment while avoiding unnecessary intervention.